Ding et al.’s Learning representation for multi-view data analysis not only provides cutting-edge research on multi-view data representation and analysis, but also provides several visual applications and practical challenges, including unbalanced or incomplete large-scale multi-view learning under what the authors call a “unified framework.”

In the introductory chapter, readers are acquainted with the problems in multi-view data analysis. The rest of the book is broadly divided into three parts. Part 1, “Unsupervised Multi-View Learning,” covers multi-view clustering with both complete and partial information (chapters 2 and 3, respectively) and multi-view outlier detection (chapter 4). Part 2, “Supervised Multi-View Classification,” covers multi-view transformation learning and zero-shot learning (chapters 5 and 6, respectively). Part 3, “Transfer Learning,” addresses the problems of missing modality transfer learning, multi-source transfer learning, deep domain adaptation, and deep domain generalization (chapters 7, 8, 9, and 10, respectively).

On the positive side, this expository book spans a good variety of research applications in the domain of big data analysis, pattern recognition, human-centered computing, web mining, computer vision, and digital marketing. On the negative side, the important and intriguing problem of masking and swamping, which is very much within the scope of the book, deserved a discussion in chapter 4 on outliers. Nevertheless, the book should be well received by advanced postgraduate students and data (especially big data) analysts. A background in statistics, mathematics, and computing is a prerequisite for reading. It is surely a must-have reference book for any scientific library.